We propose to continue a prospective study to test the hypothesis that women with low selenium intake are at increased risk of developing cancer. In addition, we will test the hypothesis that high intakes of cadmium certain other metals increase the risk of cancer by antagonizing a protective effect of selenium. Work in vitro and animal studies demonstrate that selenium supplements reduce the incidence of carcinogen-induced and spontaneous neoplasms. Epidemiologic studies relating selenium in the soil or forage crops to cancer incidence rates and our recent study utilizing prospectively collected sera support a protective effect of selenium. However, both animal studies and our human data suggest that the effect of selenium may be modified by other dietary factors. The proposed study will be conducted among the 121,964 married, female registered nurses ages 39-64 who are participants in the Nurses' Health Study (NHS), an ongoing study of determinants of cancer and cardiovascular disease using biennial questionnaires. In 1982 and 83 we collected and catalogued toenail clippings from over 68,000 members of this cohort. Keratinous structures concentrate selenium and other trace elements and we have specifically documented that levels of selenium and cadmium in the nail clippings reflect intake. Because each toe represents a different time period of keratin formation, analysis of specimens form all toes simultaneously represents intake over an extended interval. To analyze these specimens, we have employed instrumental neutron activation which is an accurate, highly sensitive, non-destructive technique. Our currently funded study will support selenium measurement of previously collected specimens from cancer cases identified by the 1984 NHS questionnaire as well as those of a series of age- matched participants who did not develop cancer. The proposed study will support the analysis of nail specimens for cancer cases and controls identified by the 1986 NHS questionnaire for selenium, cadmium, zinc, and a series of other elements. Case- control methods of data analysis will be used to address the main effects and interactions of these elements as well as the possibility that these relationships are confounded or modified by dietary or other factors for which data have previously been obtained. The size of this study will be sufficient to detect (with a power of 90% and alpha level of 0.05) relative risks of 1.5 between upper and lower quintiles of selenium for cancer at all sites, 1.7 for breast, and 2.4 for large bowel cancer.